Variability within rare cell states enables multiple paths towards drug resistance

Abstract

Molecular differences between individual cells can lead to dramatic differences in cell fate following an applied treatment, such as the difference between death versus survival of cancer cells upon receiving anti-cancer drugs. However, current strategies to retrospectively identify the cells that give rise to distinct rare behaviors and determine their distinguishing molecular characteristics remain limited. Here we describe Rewind, a methodology that combines genetic barcoding with an RNA-based readout to directly capture rare cells that give rise to cellular behaviors of interest, specifically the emergence of resistance to targeted cancer therapy. Using Rewind, we analyzed over 5 million cells to identify differences in gene expression and MAP-kinase signaling in single melanoma cells that mark a rare subpopulation of drug-naive cells (initial frequency of ~1:1000-1:10,000 cells) that ultimately gives rise to drug resistant clones. We further show that even within this rare subpopulation, molecular differences between single cells before the application of drug predict future differences in drug resistant behavior. Similarly, we show that treatments that modify the frequency of resistance can allow otherwise non-resistant cells in the drug-naive population to become resistant, and that these new populations are marked by the variable expression of distinct genes. Together, our results reveal the presence of cryptic variability that can underlie a range of distinct rare-cell phenotypic outcomes upon drug exposure. Applying Rewind to other rare biological phenomena, such as cancer metastasis, tissue regeneration, and stem cell reprogramming, may provide a means to map rare cellular states to the unique cellular fates to which they give rise.

Disrupting cellular memory to overcome drug resistance

“…We applied this technique to BRAF-mutant melanoma cells in which a rare subpopulation of cells are resistant to BRAF inhibitors. We transduced WM989 melanoma cells with a high-complexity viral barcode library consisting of a transcribed 100 base pair semi-random barcode seq ence in he 3 UTR of GFP (Emert et al 2021) (Fig. 1C).…”

“…In these examples, the cellular state underlying the phenotype is stable enough to persist through multiple cell divisions, but is ultimately not permanent, and thus amenable to switching states. For cancer therapy resistance, drug-naive cells seemingly randomly fluctuate between two states, one which is susceptible to targeted therapy and another which would become resistant if he dr g is applied, ermed primed for dr g resis ance (Emert et al 2021;Shaffer et al 2017). However, the molecular drivers underlying these heritable, but reversible, memory states remain unknown.…”

“…Recent advances in high-throughput cellular barcoding technologies have made it possible to track cellular lineages across any length of time (Bhang et al 2015). Pairing cellular barcoding with scRNA-seq enables us to now match cellular lineages with their transcriptome (Biddy et al 2018;Oren et al 2021;Weinreb et al 2020;Emert et al 2021), and thus is an ideal tool for tracking transcriptional states across lineages to measure cellular memory.…”

A transcriptionally distinct subpopulation of healthy acinar cells exhibit features of pancreatic progenitors and PDAC

“…It has been proposed, and experimentally shown, that transcriptional fluctuations within complex regulatory networks can result in a clonal population existing in multiple phenotypic states 9,10,42 . In cancer cells, such fluctuations can cause cells to switch between drug-sensitive and drug-resistant states 43 in a manner that can be perturbed by targeting key transcription factors 44 .…”

Single cell biology—a Keystone Symposia report

“…Raj showed that cells contain cryptic variability that only becomes apparent after a perturbation, like drug exposure. 64 Raj also presented unpublished data characterizing morphology, transcriptome, and phenotype across resistant cell colonies to classify resistant cells into different subtypes. Raj hopes that understanding how cells are primed for different fates will be applicable across a range of biological processes, including stem cell differentiation and induced pluripotent stem cell reprogramming.…”